DISCUSSION PAPER SERIES
IZA DP No. 3456
The Impact of Population Aging on the Labor Market:
The Case of Sri Lanka
Milan Vodopivec
Nisha Arunatilake
April 2008
Forschungsinstitut
zur Zukunft der Arbeit
Institute for the Study
of Labor
The Impact of Population Aging on the
Labor Market: The Case of Sri Lanka
Milan Vodopivec
World Bank and IZA
Nisha Arunatilake
Institute for Policy Studies, Colombo
Discussion Paper No. 3456
April 2008
IZA
P.O. Box 7240
53072 Bonn
Germany
Phone: +49-228-3894-0
Fax: +49-228-3894-180
E-mail: [email protected]
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in
this series may include views on policy, but the institute itself takes no institutional policy positions.
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center
and a place of communication between science, politics and business. IZA is an independent nonprofit
organization supported by Deutsche Post World Net. The center is associated with the University of
Bonn and offers a stimulating research environment through its international network, workshops and
conferences, data service, project support, research visits and doctoral program. IZA engages in (i)
original and internationally competitive research in all fields of labor economics, (ii) development of
policy concepts, and (iii) dissemination of research results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be
available directly from the author.
IZA Discussion Paper No. 3456
April 2008
ABSTRACT
The Impact of Population Aging on the Labor Market:
The Case of Sri Lanka*
Sri Lanka’s population is predicted to age very fast during the next 50 years, bringing a
slowdown of labor force growth and after 2030 its contraction. Based on an original, 2006
representative survey of old people in Sri Lanka conducted as a part of this study, the paper
examines labor market consequences of this process, focusing on retirement pathways and
the determinants of labor market withdrawal. The paper finds that a vast majority of Sri
Lankan old workers are engaged in the informal sector, work long hours, and are paid less
than younger workers. Moreover, as one of the first findings of its kind, the paper shows that
labor market duality that characterizes most developing countries carries over to old age: (i)
previous employment is the most important predictor of the retirement pathway; (ii) older
workers fall into two categories: civil servants and formal private sector workers, who
generally stop working before they reach 60 because they are forced to do so by mandatory
retirement regulations, and casual workers and the self-employed, who are forced to work
until very old age (or death) due to poverty and who stop working primarily because of poor
health; and (iii) the option of part-time work is used primarily by workers who held regular
jobs in their prime age employment, but not by casual workers and self-employed.
JEL Classification:
Keywords:
J11, J14, J26
population aging, labor supply of old workers, labor demand for old workers
Corresponding author:
Milan Vodopivec
The World Bank
1818 H Street, NW
Washington, D.C. 20433
USA
Email: [email protected]
*
The authors would like to Gordon Betcherman, Ralitza Dimova, Ravi Peiris, Ravi Rannan-Eliya,
Shyamali Ranaraja, Mansoora Rashid, and Nimnath Withanachchi for useful comments, and Priyanka
Jayawardena for excellent research assistance. Partial funding under the TFESSD trust fund grant is
gratefully acknowledged.
1. INTRODUCTION
In slightly more than two decades, Sri Lanka’s population will be as old as Europe’s or Japan’s
today, but its per capita income will be much lower. Indeed, while population aging is a
universal phenomenon, it looms particularly large for Sri Lanka: not only is its population among
the oldest in the non-developed world, but Sri Lanka is also one of the fastest aging countries in
the world (Figure 1). Spectacular growth would be required for Sri Lanka to catch up the
developed countries’ current aging profile at a level of per capita income these countries
currently enjoy.
In developed countries, challenges of population aging are associated primarily with the
negative impact of aging on economic growth, and the need to plan for additional public and
private outlays for old age income support and healthcare (MacKellar 2000). In Sri Lanka, a
developing country, these challenges are compounded by several facts. First, in 2006 only a
tenth of the old people received pensions as their major source of income and a third non-pension
government assistance, so most of them rely on family support as key for their wellbeing. With
only one third of the labor force presently covered by social security programs (World Bank
2007a), the need for the family support is likely to remain high in the next decades. Second,
rapid aging and modernization may put a strain on the traditional family support to old people.
The modernization of Asian societies is associated by several mechanisms that affect family
support of old people (Hermalin 2002) – lower fertility translating into fewer children available
to provide familial support; higher education levels leading to differences in attitudes and
perceptions of obligations to provide familial support; increased female labor force participation
reducing the number of caregivers available to provide support to older family members; and
rural-urban migration drawing younger persons out of rural areas to find employment.1 Third, the
health system is insufficiently focused on the healthcare needs of elderly and is constrained by
the lack of resources and their inequitable distribution (World Bank 2007b).
And fourth,
population aging may well translate into shrinking of the labor force, prompting questions of
how to promote longer working lives – and also of how to improve labor market choices for
those that are forced to continue working late in their lives.
This paper focuses on the last challenge – the impact of population aging on the labor
market. The analysis focuses on the current labor market outcomes of old workers and attempts
1
to identify both the work incentives of old people and the constraints faced by them (such as
poor health and lack of employment opportunities).2 The paper finds that a vast majority of Sri
Lankan old workers are engaged in the informal sector, work long hours, and are paid less than
younger workers. Moreover, as one of the first findings of its kind, the paper shows that duality
that characterizes labor markets of most developing countries carries over to old age: (i) previous
employment is the most important predictor of the retirement pathway; (ii) older workers fall
into two categories: civil servants and formal private sector workers, who generally stop working
before they reach 60 because they are forced to do so by mandatory retirement regulations, and
casual workers and the self-employed, who are forced to work until very old age (or death) due
to poverty and who stop working primarily because of poor health; and (iii) the option of parttime work is used primarily by workers who held regular jobs in their prime age employment,
but not by casual workers and self-employed.
The paper is organized as follows. To set the stage, Section 2 presents long term labor
force projections and discusses economic implications of the slowdown of labor force growth
and its shrinking. The next sections analyze labor market outcomes of old workers: Section 3
describes key employment and pay outcomes for old workers and the patterns of their
withdrawal from working life; Section 4 examines pathways to retirement, that is, typical ways
in which workers transit from full-time employment to complete retirement; and Section 5
explores the determinants of work activity of old workers with multinomial logit models. Section
6 discusses possible obstacles for employment in old age, and Section 7 concludes with policy
implications.
2. LABOR SUPPLY AND GROWTH IMPLICATIONS OF POPULATION AGING
Sri Lanka’s population is set to age rapidly during the next 50 years, with the proportion of old
people increasing substantially. The main drivers of this process are low and declining fertility
and increasing life expectancy. The total fertility rate (the number of children that the average
woman bears during her lifetime) fell below the replacement level of 2.1 by 1994 and it has
continued to fall, reaching 1.7-1.9 in the 2000s. Increasing life expectancy is another driver of
Sri Lanka’s population aging. Currently life expectancy for men is 69 years and for women is 77
years, higher than in most developing countries, and present trends indicate that life expectancy
2
will reach the current average OECD level of 77.8 years by 2050 (OECD, 2005). Central to Sri
Lanka’s demographic aging will therefore be an increase in the share of the old people, with the
share of peopled aged 60 years and above projected to increase from current 11 percent to 16
percent in 2020 and to 29 percent in 2050 (Figure 1). According to De Silva (2007), the share of
very old people (80 years and more) will also strongly increase. Because in the next 2-3 decades
the increasing old age dependency ratio (the proportion of population aged 60 years or more
versus the proportion aged 15-59 years) will be roughly matched by the fall in the child
dependency ratio (proportion of population aged less than 15 years versus the proportion aged
15-59 years), the overall dependency ratio is projected to start increasing rapidly after 2035
(Figure 2).
Population aging will translate into a slowdown of labor force growth and later into its
contraction, and its composition will also be affected. Labor force projections show that Sri
Lankan labor force will continue to grow for another two decades, and it will shrink thereafter in
a foreseeable future.3 Applying constant current labor force participation rates to population
projections, Figure 3 shows that labor force will stop growing around 2030 (in 2027, 2032, and
2037 for the low, standard, and high population scenario, respectively), and will thereafter start
to shrink, dropping to the current size of the labor force in about 30 years under the standard
population scenario.4 Population aging will also significantly change the age composition of the
labor force, with the share of workers younger than 30 years significantly shrinking and the share
of those older than 50 years strongly increasing (Figure 4).
With labor force growth being one of the sources of economic growth, the slowdown of
labor force growth and its shrinking after 2030 may contribute to a slowdown in GDP growth.
While simulations that would determine the aging-induced “tax” on GDP growth is beyond the
scope of this paper, it is instructive to present some estimates of this tax for other countries.
Because of population aging, the forecast of GDP per capita growth in the OECD area is reduced
to around 1.7 percent per year over the next three decades, bringing about 30 percent reduction in
comparison to the 1970-2000 growth rate (OECD 2006). Martins et al (2005) estimate that in the
next half century, the decline of their workforces will contribute to a reduction of the per annum
growth rates by 0.2 to 0.5 percentage points in France and Germany, and by 0.8 percentage
points in Japan. Moreover, because of the raising share of inactive population, the economic
growth in Japan, France and Germany will slow down by additional 0.2 to 0.3 percentage points,
3
compared to growth under a stable age structure of the population. For comparison, about one
quarter of the growth of OECD countries in recent decades has been attributed to labor force
growth (Martins et al 2005).
Population aging, however, may not necessarily reduce economic growth, if
countervailing factors prevail. A drag on economic growth created by population aging could be
countered by higher overall (multifactor) productivity, increases of labor quality, or increased
labor force participation. While it is hard to predict the changes of multifactor productivity,
recent trends show that education level of workforce has improved significantly in the recent 15
years (the share of workers with finished A level or above has increased by 5 percentage points,
and with commensurate reduction of workers with less than 5 years of schooling), and the quality
of workforce is likely to improve further in the future. Another plausible countervailing factor is
the increase of labor force participation rates – for example, simulations show that the increase of
women’s participation rate by 20 percent could delay the reduction of labor force by 15 years.5
3. WORKING IN OLD AGE
Below we describe key labor market outcomes for old workers, focusing on the timing of
retirement, work load, manner of retirement (how quickly people transit from full-time
employment to complete retirement and the use of part-time work before reaching complete
retirement), changes in the sector of work and occupation, and reasons for retirement.
Labor force participation and unemployment of old people
During 1992–2004, labor force participation of old workers remained stable, men’s participation
rate by far exceeded women’s, and the withdrawal from a labor market occurred rather late in the
life cycle (Figure 5). The only exception in regard with stable trends was a rising LFP rate of
females aged 50-59, increasing from 26 percent in 1992 to 33 in 2004; in contrast, LFP for males
in that age show no pronounced trend and remained in the 51-56 percent range (Figure 6).
Workers were withdrawing from the labor market rather late in the working career – starting
with the age group 60-69, for males, and with the 50-59, for females (Table 1). For males, LFP
rates for all age groups in the prime age (20-59) exceeded 80 percent, dropping to the 50s in the
age group 60-69 and to around 20 percent in the age group over 70. Withdrawal for the women
4
starts somewhat earlier – in 2004, LFP rates in the age group 40-49 were about 44 percent,
dropping to 33 percent in the age group 50-59, to 14 percent in the age group 60-69, and to 3
percent in the age group over 70.
Unemployment among old Sri Lankans is a rather uncommon phenomenon. During
1992-2004, unemployment rates for males aged 60-69 were typically well below 1 percent and
virtually 0 for those over 70 years; unemployment rates for old women were even lower (Table
2). These figures reflect both the fact that when losing their jobs, old workers tend to withdraw
altogether from the labor market (the phenomenon called “hidden unemployment” by OECD
2006). Unemployment rates are rapidly falling with age for both men and women (Figure 7). In
2004, 15-19 year olds had the highest unemployment rates, amounting to 26 and 37 percent,
respectively, for men and women, with unemployment rates for older age groups monotonically
decreasing and reaching single digit numbers already in the age group of 30-39 year olds.
More rigorous analysis confirms strong association of labor force participation and
unemployment with age. The above associations were obtained from cross-sectional data and not
from observing the same set of individuals through time, and changes in the composition of
persons observed in different age categories may influence the results. Accounting for such
differences, probit regressions, run separately for males and females, confirm that age has a
strong influence on the labor force participation and unemployment of individuals.6 Relative to
labor force participation of individuals aged 20-24, the marginal effect of age is very strong in
the late 20s, remaining flat till about the late 40s, and becoming increasingly negative for both
males and females (not shown). Similar probit regressions also confirm the association of age
with unemployment rates, with age accounting for a much larger reduction of unemployment rate
between early 20s and late 30s for women than for men.
International comparisons show that Sri Lanka’s labor force participation rates for old
people lag behind the rates of its regional comparators while mostly exceeding those in
developed countries. Labor force participation of older men is lower than in other countries in
the region (India, Thailand, Bangladesh and Pakistan, Table 3), and so is participation of women,
except that women’s participation rate in Sri Lanka exceeds the one in predominantly Muslim
countries (Pakistan and Bangladesh). Interestingly, the LFPs of elderly in Sri Lanka also lag
behind the rates of Japan, but they exceed other developed countries.
5
Unemployment rates of elderly Sri Lankans are in line with those of regional comparators
and lag substantially behind developed countries. Unemployment rates for elderly – both males
and females – in Sri Lanka are similar to those in India and Thailand, where open unemployment
among old workers is also virtually unknown. Reflecting a different institutional set-up – above
all, the existence of specialized programs providing cash benefits to unemployed that stimulate
labor force attachment – unemployment rates of old workers in developed countries are
significantly higher than in developing countries (although they are still much lower than youth
unemployment rates in these countries). In 2004, unemployment rates for workers aged 50-64
ranged from 2 percent in Norway to 11 percent in Germany (OECD 2006).
Employment status, work load and pay of old workers
Most of the old workers have full-time rather than part-time jobs. As shown above, as they age,
fewer and fewer workers remain employed. But if they do, they tend to work in full-time jobs
rather than part time. In our survey of old people, 30 percent of males were employed, of whom
two thirds were in full time employment and only a third in part-time employment; only 5
percent of women were employed, with 60 percent of them in full-time employment. The
proportion of part-time employment among old workers was the lowest in the 60-64 age group
and then increased for older groups of workers, with the share of part time workers exceeding the
share of full-time workers in the 75-79 age group (see Figure 8).7
The most prevalent way of withdrawing from the labor market is “overnight,” that is,
without engaging in part-time employment. As implied by falling LFPs and the fact that parttime work is relatively infrequent, most workers exit from full-time employment directly to
complete retirement. In our sample of individuals who held stable jobs in their early fifties (see
selection details below), a vast majority (89 percent) of workers who did not work full time any
longer had retired "overnight," that is, they did not work part time before retirement and they
completed transition to inactivity in a very short time.
The number of hours worked is also reduced by age. Reflecting retirement patterns, old
workers as a group are providing less and less labor to the economy. The average weekly number
of hours of work provided by old male workers decreased from 23 for 60-63 year olds to below
10 for the 72-75 year olds; and by old female workers, from 5 and 2 for the same are groups
(Figure 9a). At the same time, workers staying active reduce the number of hours worked as they
6
age, men from the average of 47 in their early sixties to 36 in the 72-75 year group, and women
from 35 to 29 for the same age groups (Figure 9b). Despite these reductions, the average
numbers of work for those workers who remain employed remain large even in high age. Old
workers in part time jobs also work long hours – in our sample (SLAS 2006), they on average
work only 12 hours less than old workers in full-time jobs (36 hours compared to 48).
In comparison to younger workers, old worker are paid less, particularly in the public
sector. Wages of workers of 65 years and above are only a fraction of wages of the best paid
group of workers – workers in their late 50s and early 60s, in the public sector, and in their 30s,
in the private sector (Figure 10). Particularly strong reduction of wages of older workers occurs
in the public sector, with workers aged 65 and over earning barely over 1/3 of what the workers
in the 60-65 group are earning, and only about 20 percent of what workers in the 55-59 year
group are earning (the latter applies for women only). In the private sector, the reduction of
wages for workers above 65 years of age is more modest. Interestingly, except for women in the
public sector, wages of workers in the 60-64 year group are higher than wages of the 55-59 year
group.8 One salient feature of the age-wage profiles is the fact that the both men’s and women’s
profiles of the private sector are much flatter than the ones in the public sector. Many OECD
countries have hump-shaped age wage profiles, but there are many exceptions, reflecting
different institutional wage-setting and other arrangements (see OECD 2006).
Where and as what do elderly work?
A vast majority of old workers are self-employed or casual workers engaged in the informal
sector. More than half of the working elderly males and females are self-employed (Figure 11a).
Majority of the rest of workers are casual workers, with less than 3 percent of elderly working as
regular workers, mostly workers in their early sixties. Vast majority of old workers are engaged
in informal private sector (the main exception being men in their early sixties, with small share
of them working for the government or in the formal private sector) – see Figure 11b.
Old workers, both males and females, are working mostly as skilled workers in
agriculture, manufacturing, and wholesale and retail trade (Figure 12). A fair proportion of them,
particularly of women, are working also at the bottom of occupational ladder (in elementary
occupations), and a small proportion as professional workers. Not surprisingly, the composition
of workers by occupation changes little with age, as does sector of work, except that the share of
7
workers in agriculture is reduced. While sector of work and status for old workers working parttime does not differ much from full-time workers, a larger share of them are professional and
skilled workers.
Reasons for retirement and part-time work
The main reasons for retirement are on the push side: ill-health and reaching mandatory
retirement age, for both males and females. 9 Nearly half of retired old workers stated that health
reasons were very important for their decision to retire, and about 40 percent of them quoted
reaching the mandatory retirement as being the second most important reason (Table 4). Among
the push factors, work and travel stress, as well as the closure of businesses were also important
factors affecting the decision to retire.
Among the pull factors, “doing other things (hobbies)” was an important reason for
retirement, especially for females (Table 4). Only females found family obligations a very
important reason for complete retirement; however, it was a relevant factor only for a small
proportion of females. This indicates that obligations of child-care and care for old persons do
not have a large effect on retirement decisions.
Earning additional income was the main reason for part-time employment for most
workers. Three in four elderly in part-time employment are continuing to work to earn additional
income. More females than males consider the additional income from part-time employment as
the main reason for continued employment. A small percent also worked part time to keep them
selves occupied (6 percent) and to be in touch with their profession (3 percent). A large share of
elderly working part-time does so involuntarily, as 80 percent of those who retired after doing
some part-time work indicated that they would have liked to stop work completely after
retirement from full-time work. (On the other hand, over a quarter of the surveyed old people
indicated that they would have liked to continue doing some paid work when they retired from
their full time jobs.)
The reasons for retirement differ substantially from those in OECD countries. Similar to
workers in OECD countries, a share of workers retires upon reaching mandatory retirement age.
However, unlike in OECD countries, the single most important reason for withdrawal from the
labor market is poor health. A smaller share of workers is also pulled into retirement by family
obligations, work stress, travel stress and work dislike. Moreover, in OECD countries there are
8
many institutional arrangements that entice old workers to retirement or early retirement, ranging
from pre-retirement pension schemes to disability and unemployment insurance benefits (OECD
2006) – the arrangements largely missing in Sri Lanka.
4. PATHWAYS TO RETIREMENT
Above we analyzed the patterns of withdrawal from working life as people age, treating the
population of old people as a whole. Remembering a key motivation for this investigation –
identifying ways to promote choice for old workers in both improving job prospects as well as
reducing pressures on those who are forced to work – it is helpful to find out whether the
working careers themselves determine the fate of workers in their old age. Recognizing the deep
segmentation of the Sri Lankan labor market, we do so by tracing retirement paths – typical
sequences of labor market arrangements between full-time work late in the working career and a
complete retirement/withdrawal from active working life – for different groups along these
segmentation lines. We therefore focus on the following four groups of workers:10
y
Regular public sector workers (i.e., regular workers of public sector, public corporations or
boards and the co-operative sector),
y
Regular private sector workers (i.e., regular workers of both the formal and the informal
private sector),
y
Casual workers (i.e., casual and contractual workers), and
y
Self-employed (i.e., employers and the self-employed).
In our analysis of retirement pathways and exploiting the data from the SLAS 2006, we focus
on the following aspects (the most relevant ones among the aspects studied in the previous
subsection): timing of retirement, manner of retirement (“overnight” vs. via part-time
employment), changes of sectors of work and occupation, and reasons for retirement. Note that
for this analysis we exploited a panel nature of the SLAS 2006 – the fact that in addition to being
asked asking about their current employment status, respondents were retrospectively asked also
about the age of retirement from their full- and part-time job, as well as their employment status
when whey were in their early fifties, that is, at the end of their prime-time work career. Based
on this information, we selected 1060 workers with stable jobs in their early fifties (they held
9
full-time jobs and were continuously working when they were 51-54 years old), and traced their
work history continuously into their old age.11
Our results confirm that pathways to retirement show strong dual patterns. The majority
of regular workers in public and private sector retire early, most of them before they reach 60,
and for work-related reasons (including mandatory retirement); in contrast, a large share of selfemployed and casual workers continues to work full-time into very old ages, and most of them
withdraw from employment for health reasons. The option of part-time work is also used
primarily by workers who held regular jobs in their prime age employment, but not by casual
workers and self-employed. The results show that labor market duality, the hallmark of labor
market in many developing countries, carries over to old age and importantly determines the fate
of old people. We present detailed results below.
Timing of retirement: regular public and private sector workers withdraw from the labor
market much earlier than casual and self-employed workers. More than two thirds of workers
holding regular jobs in the public sector in their prime age, and 57 percent of workers holding
jobs in the private sector, completely retired when they were 60 years old. In contrast, only 23
and 18 percent of workers who were casual workers or self-employed in their prime age retired
when they reached the age of 60, with the gap in retirement status widening with the age of
workers (Figure 13)! When they reach 69, virtually none of the workers who have spent their
careers in the formal sector is still working full time; in contrast, nearly half (47 percent) of their
counterparts who were casual workers and self-employed are still working full time.12 Note that
in terms of occupations, the first two categories comprise mostly of white-collar workers
(officials, managers, clerks and professional), in contrast to predominantly blue-collar workers
comprising the casual and self-employed (who were mostly skilled agriculture and fishery
workers, craft and related workers, and workers in elementary occupations). The observed
pattern of retirement is thus contrary to the one found for OECD countries, where blue-collar
workers and less-skilled workers are more likely to retire earlier (see OECD, 2006).
Another way of looking at differences at retirement patterns is to observe retirement ages.
Figure 14 shows retirement from full-time employment for the selected sample of individuals
who held stable jobs in their early fifties described above. The distribution of retirement age is
similar for regular public and private sector workers, with the 95 percent of workers from the
first group and 84 percent from the second group retiring between ages 55 and 64 (for public
10
sector workers, there are clear peaks at ages 55 and 60, due to mandatory retirement – not shown
on the graph). In contrast, the distribution of retirement age for informal sector workers spreads
in much higher age, with 16 and 18 percent of casual and self-employed workers, respectively,
retiring when aged 76-80.
Manner of retirement: the majority of workers across all groups retire overnight, with
labor market withdrawal via part-time employment being more common for regular workers. In
our sample of individuals who held stable jobs in their early fifties, 11 to 13 percent of regular
public or private sector workers held part-time jobs by the time of the survey, and only 6 percent
of casual and self-employed workers did. (Note that at the time of the survey, around 40 percent
of casual and self-employed still held full time jobs and very few among the regular workers.)
Industry and occupational composition: changes occur primarily for regular but not
for casual and self-employed workers. Proportions of elderly in different industrial sectors
remain constant for casual and self-employed workers (see Figure 15). In contrast, among
workers who held regular jobs at the end of their prime-age working career, a proportion in
agriculture increases with age (after age of 67, workers with jobs in public sector tend to
withdraw from the agriculture and their share in industrial sector increases).
Similarly,
occupational composition for casual and self-employed workers shows little change with time,
while among workers who held regular jobs at the end of their prime-age working career, the
share of skilled workers increases and the share of professional workers decreases with age (not
shown).
Reasons for retirement: most workers are pushed away from jobs, but while for regular
public and private sector workers the main reason is mandatory retirement age, for casual and
self-employed workers is ill-health. For regular public sector workers, the main reason for
retirement at any age is work related, primarily reaching mandatory retirement age (see Figure
16). The same holds true for regular private sector workers, except for those aged 61-70. Overall,
a slightly over a quarter of regular workers in the private sector cited reaching mandatory
retirement age as a reason for retirement from full time work. On the other hand, by far the most
compelling reason for retirement at all ages for casual workers and self employed was ill health
(a small exception are casual workers aged 55-59, where work-related reasons slightly
overweigh health reasons). Interestingly, personal reasons figured more prominently for the
groups of casual workers and self employed than for regular workers.
11
5. DETERMINANTS OF RETIREMENT
The above analysis is mainly descriptive, examining the retirement from active working life
through a “single lens,” that is, one aspect at the time (section 3), and contrasting retirement
patterns across workers with different employment status during their working life (section 4). In
this section, we extend the approach to multivariate analysis, that is, to the identification of the
separate influence of individual variables on retirement decisions, the influence that may be
clouded in a simple, univariate analysis.
Speaking formally, below we report the results of multinomial logit models identifying
factors that contribute to the withdrawal from the labor market. The dependent variable is the
change of the labor market status of individuals holding stable, full-time jobs at age 54 years old
as compared to their status when they were 61-65 year old (1060 persons in the SLAS).
According to work status during ages 61 to 65, we allow three possibilities for this change (that
is, our dependent variable can take three values):
y
Completely retiring upon reaching 61 years and not working when aged 61-65 (in our
sample, 382 individuals); this transition was taken as a baseline.
y
Working at least some time during ages 61-65, either part time, or full-time if not throughout
the period (in our sample, 208 observations).
y
Working continuously full-time during ages 61-65 (in our sample, 470 observations).
Among explanatory variables, labor market factors, as well as family, demographic and
personal characteristics are included (see Table 5 for descriptive statistics of independent
variables). Of the sources of income considered in the study, inclusion of financial or in-kind
family assistance could be endogenous, if one believes that the amount of help given by family
vary according to other sources of income, and health status of elderly. Hence, instead of using
family assistance, we use a proxy variable, which indicates whether individuals have access to
help if they need it. Statistically significant results of this analysis are highlighted below.
Employment status at age 55 is the most important factor determining work activity of
old workers. The results show that keeping other things constant, compared to regular public
sector workers, all other types of workers are more likely to be working full-time during ages
61–65 (Table 6). These effects are large for workers of all three groups – regular private sector
12
workers, and particularly for casual and self-employed workers: workers from the first group are
40 percent more likely to work in a full-time job in their early 60s than workers who left regular
public sector jobs, and casual and self-employed workers 66 and 69 percent more likely,
respectively. Interestingly, while casual and self-employed workers are more likely to work full
time, but they are less likely to work part time as compared to workers who were regular public
sector workers at the end of their working career. In other words, while the option of working
part time is open to previous regular public sector workers, this option seems to be less available
to casual and self-employed workers.
The probability of working in old age decreases progressively with (self-perceived)
health problems and the presence of chronic illnesses. The results show that relative to
individuals who reported very good health, the probability of full-time or part-time work of
individuals who reported good or poor health is significantly smaller – for those in poor health,
the chances for full-time work are 23 percent smaller, and for part time work 4 percent smaller,
in comparison to those in very good health. Similarly, the probability of full-time (but not parttime) work of individuals who reported a chronic illness is also significantly smaller (by 11
percent) in comparison to individuals who did not report a chronic illness.
Pension income is a major factor reducing labor supply of elderly. The results of the
multinomial model show that pension income has a large and negative effect on work activity in
old age. Receiving a pension reduces the probability of full-time work by 21 percent and that of
part-time work by 7 percent. Location also affects elderly work activity, but the effects are small.
Estate sector workers are less likely to work full-time during ages 61-65, relative to urban sector
workers. This could possibly be due to the high level of strenuous physical activity associated
with estate sector work.
Interestingly, individual characteristics – gender, ethnicity, marital status, being a head of
a household – have no significant effect of the studied labor market transitions. We do find,
however, that men are more likely then women to make a transition to part-time jobs, as do
previously married workers as compared to currently married.
13
6. OBSTACLES FOR EMPLOYMENT OF OLD WORKERS
A wide range of outcomes in the Sri Lankan labor market is suggestive of a presence of rigidities
regarding employment conditions faced by old workers. We showed that old workers long
relatively long working hours – according to the 2006 SLAS, only 12 percent of old workers in
part-time employment and 8 percent in full time employment were able to adjust their working
hours. Moreover, only 8 percent of retired workers had the option of reducing the number of
working hours in their last full-time job, and about a quarter of the old workers who retired
overnight would have preferred to do some paid work before completely retiring. Other findings
of the analysis above complement this picture: we showed that the main reason for retirement of
many regular workers was mandatory retirement, that only a small part of workers take part-time
employment jobs, and that LFP rates of old workers in Sri Lanka are lower than in regional
comparators. Below we probe into the possible obstacles for employment of old workers –
inflexible employment legislation, high relative wages of old workers, and work stress and weak
employability of old workers.
Inflexible employment legislation
While lack of adjustment of working hours, low LFP rates, and the fact that people are retiring
despite wishing to continue working in paid jobs are suggestive of inflexibilities, there may be
other reasons for apparent rigidities in the labor market outcomes of old workers. Employers
may be reluctant to hire old workers on part-time basis because doing so provokes difficulties of
coordinating work in teams; it may generate demands of other groups of workers to offer
reduced working hours to them as well; and, possibly, runs against outdated attitudes of
employers about part-time employment. Note that the obligation to pay gratuity does not work
against old workers, because their relative pay is low (see the discussion of age wage profiles
above).
Sri Lanka’s strict employment protection legislation, embodied in the 1971 TEWA
system, however, may be a source of disincentives for employment of old workers.13 In
principle, it is not possible to predict the impact of strict employment protection legislation
(EPL) on labor market outcomes for old workers, as it may lead to both greater retention of old
workers as well as to the reduction in their hiring. While the empirical evidence is weak, some
recent studies report negative effects of strict EPL on employment of old workers. For example,
14
a cross-section analysis of OECD countries shows a statistically significant association of strict
EPL with reduction of employment and hiring rate of workers aged 50-64 (OECD 2006). There
is also a number of country studies which identified strict EPL as a barrier to the hiring or
retention of older workers.
There is another channel through strict employment protection legislation affects
employment of old workers – it encourages the use of mandatory retirement stipulations in
employment contracts. While in Sri Lanka there is no legally binding retirement age (except for
civil servants), employment contracts typically do specify the retirement date. The reason is
simple: similar to other countries with large separation costs, setting the retirement data avoids
costly layoffs.14 Because Sri Lankan employers can extend employment contracts of old workers
beyond pre-set retirement dates, or hire old workers who retired from other firms, without being
liable to severance payments (except gratuity), the necessary flexibility in employing old
workers is achieved. Increasing the mandatory retirement age for civil servants, set currently at
55, would probably help keeping these workers employed for a longer time, but there is a host of
other consideration that should be taken into account (above all, setting consistent hiring
practices and considering revisions of wage scales) for this measure to be useful rather than just
to increase the costs of bureaucracy.
Are wages of old workers set too high?
Setting the level of wages too high could work against the employment of old workers. More
precisely, if wage increases associated with age exceed the increases in worker productivity,
employers would prefer hiring young workers and would be reluctant to retain old workers.15
Indeed, in many countries wages by age show a hump-shaped profile, the profile that is most
consistent with high level of employability of old workers (see OECD 2006 for a description of
such profiles in OECD countries).
The hump-shape profile could be explained by rising
productivity of workers as they gain experience and falling productivity after a certain age.
As shown above, in Sri Lanka relative costs of old vs. young workers appear not to create
employment disincentives. The age wage profile for the private sector reaches its peak for the
30-34 age group, slowly decreases thereafter, and experiences a sharp drop after the age of 60.
While this drop may reflect the selection process that is not accounted for in age-wage profiles
based on cross-section data, the magnitude of the drop of wages of old workers nonetheless
15
suggests that the level of wages of old workers – and thus the wage-setting mechanism – is not a
disincentive to hire or retain old workers in Sri Lanka.
Other factors
Work stress and stress of traveling to work may also create obstacles to employment of old
workers. As shown above, a fair share of elderly have stopped work due to work stress and the
stress of traveling to work. Work stress may be related to many factors, including inappropriate
working conditions, discrimination of old workers, or the perception of the old workers about
losing their ability to cater the needs of the market. OECD reports that large shares of old
workers in OECD countries is exposed to physically unpleasant conditions (43 percent of men
and 26 percent of women aged 50-64), and the conditions in Sri Lanka are probably no better.
Another likely obstacle for employment of old workers may be their outdated skills and
negative perceptions of employers about the adaptability and productivity of older workers. Even
in OECD countries, improving training opportunities and the willingness to take advantage of
them have been identified as a key area for strengthening employability of old workers. More
research is needed to determine the needs – and the likely impact – of such training in Sri Lanka.
And to change negative employer attitudes, a voluntary approach fostering public information
campaigns and guidelines about employment of elderly may help.
7. CONCLUDING REMARKS
The paper showed that in the coming decades, population aging in Sri Lanka will bring a
slowdown of labor force growth and after 2030 its contraction. By studying labor market
outcomes of old workers, the paper also demonstrated that labor market duality carries over to
old age: while civil servants and formal private sector workers stop working early (generally
before they reach 60 years) because they are forced to do so by mandatory retirement provisions,
casual workers and the self-employed, in contrast, work until very old age (or death) because
they are poor and have to earn their living, and stop working when they are forced to do so for
health reasons.
How should labor market and other policies respond to such circumstances? To counter
the prospects of labor force reduction, policies could be aimed at increasing participation rates,
16
particularly of women and older workers. To increase labor force participation of women, the
introduction of family-friendly policies (including increasing part-time working opportunities)
would be helpful, as would complementary policies aimed at reducing their burden as primary
care givers to old persons (see World Bank 2007b). To increase labor force participation of old
workers, retirement age should be made more flexible. More research is needed to examine other
possible obstacles for employment of old workers (such as inflexible labor laws) and to
determine whether employers’ negative perceptions about the adaptability and productivity of
older workers create work disincentives for old workers – and if so, what should be done about
it.
Another area is improving productivity of the labor force. One important task is to
improve skills of older workers through investments in life-long learning. While it is true that
older workers undergo training less frequently, increasing the expected duration of their working
lives will help them to be more motivated to take training – and it will also increase the payoffs
of training to employers. Moreover, OECD (1998) argues that if workers undergo continued
training throughout the career, it is unlikely that they would experience a decline in their
trainability in the old age. In the context of developing country, the second important route to
make labor more productive is formalization of the economy. On this front, one key task is
making the labor market legislation less restrictive, so that workers could shift towards better,
more productive jobs – and, at the same time, jobs that offer improved social security. And there
is also a third channel of raising productivity: improving health outcomes of old workers, the
channel that contributes both to higher earning capacity and to longer working careers of
workers.
The last point underscores the importance of improving the choices available to old
workers. As documented above, many workers – including a vast majority of informal sector
workers – stop working only when they are prevented from doing so by ill health, so improving
the health status of old people would significantly relieve financial pressures on them. Old
peoples’ choices would also significantly improve by providing them with an independent source
of income, which underscores the need to extend the coverage of old age income support
systems.16
In a country where extending working lives is key to relieve workers of financial
pressures, our finding that ill health – more precisely, according to World Bank 2007b, chronic
17
illness – is a major cause of withdrawal from the labor market, deserves special attention. While
it highlights the need to formulate a coherent policy for addressing the health challenges of
aging, it also points to the special needs of low socioeconomic groups, because these groups, as
shown by this study, are most likely to be forced to work long into their old age. Indeed, World
Bank (2007b) shows that in Sri Lanka, the burden of ill-health from untreated noncommunicable diseases (NCDs) affects the poor more than the non-poor: the poor have worse
access to medical services and so they are less likely than the non-poor to have their NCDs
detected and treated. But the need to re-focus health policies to cater better to the needs of old
people is clear also from the fact that physical disability rates among the old people in Sri Lanka
in the past two decades have increased, in contrast to many other health outcomes.
18
References
Hermalin, A. I. 1995.”Ageing in Asia: Setting the Research Foundation.” Asia-Pacific Research
Reports, No. 4. East-West Center, Program on Population.
Hermalin, A. I. 2002. The Well-Being of the Old Persons in Asia: A Four-Country Comparative
Study, Ann Arbor: University of Michigan Press.
Hermalin, A. I.; and Yang, L.S. 2004. “Levels of Support from Children in Taiwan: Expectations
versus Reality, 1965-99.” Population and Development Review, 30 (3), pp. 417-48.
Institute for Health Policy. 2007. “Socio-Economic and Health Profile of Elderly in Sri Lanka.”
Background paper prepared for the project on population aging in Sri Lanka, processed,
Institute for Health Policy, Colombo.
Kim, I. K. 1999. “Population Ageing in Korea: Social Problems and Solutions.” Journal of
Sociology and Social Welfare 26(1): 107-123.
Knodel, J. and M. B. Ofstedal. 2002. “Patterns and Determinants of Living Arrangements.” In
A. I. Hermalin (ed). The Well-Being of the Old persons in Asia. University of Michigan
Press.
MacKellar F. L. 2000. “The Predicament of Population Aging: A Review Essay,” Population
and Development Review, June 2000, v. 26, iss. 2, pp. 365-97.
Martins, J.O., F. Gonand, P. Antolin, C. de la Maisonneuve, and K.Y Yoo. 2005. “The Impact of
Aging on Demand, Factor Markets and Growth,” OECD, Economics Working Papers
No. 420.
Muenz, R. 2007. “Aging and Demographic Change in European Societies: Main Trends and
Alternative Policy Options.” Social Protection Discussion Paper No. 03/07. Washington
D.C.: The World Bank.OECD. 2005. Health at Glance: OECD Indicators. Paris, France:
OECD.
OECD. 2006. Live Longer, Work Longer. Paris, France: OECD.
de Silva, W. I. 2007. “A Population Projection of Sri Lanka for the New Millenium, 2001-2101:
Trends and Implications.” Colombo, Sri Lanka: Institute for Health Policy.
World Bank. 1994. Averting the Old Age Crisis. Policies to Protect the Old and Promote
Growth. Washington, DC: The World Bank.
World Bank. 2007a. Sri Lanka: Strengthening Social Protection, World Bank: Human
Development Unit, Report No. 38197-LK.
19
World Bank. 2007b. Sri Lanka: Addressing the Needs of an Aging Population. World Bank:
Human Development Unit, South Asia Region (forthcoming).
20
Table 1: Labor force participation by age group from 1992-2004 (%)
(a) Males
Year
15-19
1992
32.1
1993
33.8
1994
34.0
1995
31.2
1996
31.0
1997
31.0
1998
35.4
1999
36.7
2000
32.9
2001
28.8
2002
30.1
2003
29.3
2004
29.5
(b) Females
Year
15-19
1992
21.1
1993
22.8
1994
21.2
1995
19.8
1996
20.1
1997
20.1
1998
23.0
1999
21.3
2000
18.2
2001
17.2
2002
19.0
2003
18.6
2004
18.1
20-29
89.3
88.7
87.9
87.7
88.6
88.6
89.6
89.1
87.7
88.1
88.5
87.9
88.1
30-39
96.7
96.9
96.0
96.3
95.8
95.8
97.0
96.7
96.6
96.0
96.9
96.4
96.3
40-49
95.0
94.6
95.8
94.5
94.7
94.7
95.5
95.0
95.6
95.6
95.3
94.8
94.7
50-59
80.1
78.5
82.7
82.0
82.5
82.5
84.2
84.9
84.2
84.8
85.4
83.6
83.9
60-69
54.9
56.0
53.6
51.4
52.5
52.5
56.3
53.5
54.3
52.6
54.8
53.4
51.9
over 70
21.4
25.6
23.1
22.5
23.8
23.8
23.1
25.7
22.1
21.2
23.6
22.4
18.2
Total
75.2
75.6
76.4
74.8
75.4
75.4
77.0
76.7
76.1
75.2
76.7
76.3
76.0
20-29
52.5
54.2
53.2
51.7
49.9
49.9
54.8
50.6
50.1
49.0
50.0
44.5
45.4
30-39
45.6
48.7
47.4
46.0
45.4
45.4
51.3
48.7
50.4
44.6
45.4
44.0
43.7
40-49
40.8
45.9
41.2
44.3
43.6
43.6
50.9
48.0
47.3
45.1
47.2
44.1
44.4
50-59
25.8
27.1
27.5
28.2
29.6
29.6
34.8
34.5
34.8
34.0
35.6
34.3
33.1
60-69
11.5
11.4
11.6
14.0
14.5
14.5
17.8
15.4
15.6
15.2
13.5
13.7
14.0
over 70
3.6
3.4
3.4
3.8
3.2
3.2
5.1
3.4
4.2
3.3
5.0
5.0
3.4
Total
36.2
38.2
36.7
36.3
36.1
36.1
40.9
38.2
37.9
35.7
37.4
35.1
35.0
21
Table 2: Unemployment rate by age group from 1992-2004 (%)
(a) Males
Year
15-19
1992
38.5
1993
34.5
1994
35.5
1995
31.7
1996
30.1
1997
30.1
1998
25.5
1999
25.5
2000
22.1
2001
28.9
2002
26.9
2003
27.1
2004
26.3
(b) Females
Year
15-19
1992
46.4
1993
44.9
1994
49.6
1995
48.4
1996
39.6
1997
39.6
1998
30.6
1999
34.1
2000
26.3
2001
31.4
2002
35.4
2003
35.6
2004
37.4
20-29
20.4
18.2
18.4
18.4
16.9
16.9
13.7
14.3
13.5
14.3
15.7
14.3
14.3
30-39
4.9
4.9
5.0
4.7
3.5
3.5
2.5
2.5
2.2
1.8
2.2
2.0
2.3
40-49
2.2
2.3
2.6
2.8
1.0
1.0
1.3
1.5
0.8
1.3
1.4
1.2
1.4
50-59
1.8
2.3
1.9
0.6
0.7
0.7
0.9
1.6
0.8
0.7
1.0
0.8
0.8
60-69
1.1
0.2
0.6
0.1
0.7
0.7
0.9
0.5
0.2
0.1
0.3
0.4
1.1
over 70
0.8
0.2
0.2
0.0
0.0
0.0
0.0
0.4
0.0
0.0
0.0
0.0
0.6
Total
10.7
9.6
9.6
9.0
7.6
7.6
6.5
6.8
5.8
6.2
6.6
6.0
6.2
20-29
39.4
36.6
34.6
33.6
32.0
32.0
28.2
27.0
24.3
25.9
27.2
29.1
29.4
30-39
12.9
15.1
12.5
12.7
8.3
8.3
9.1
8.0
6.3
6.5
7.2
7.5
6.3
40-49
4.2
5.3
2.6
2.9
2.8
2.8
3.3
1.7
2.3
1.6
1.7
2.4
2.4
50-59
0.8
1.9
0.2
0.3
0.7
0.7
0.6
0.9
1.6
0.4
1.1
0.6
0.7
60-69
1.4
0.0
1.3
0.0
0.4
0.4
0.0
0.2
0.0
0.0
1.2
0.4
0.8
over 70
0.0
2.1
1.0
0.0
0.0
0.0
0.0
2.2
0.0
1.5
1.9
0.0
0.0
Total
22.4
21.7
20.1
18.9
16.2
16.2
14.2
13.2
11.1
11.5
13.0
13.2
13.1
22
Table 3: International comparisons of labor force participation rates for elderly
Age grouSri LanIndia Bangla-dPakistan Thai-lanPhilip-Japan Hong US
pines
Kong
France Germany Norway Switzerland
Year of da2004
2001 2003
2004
2004
92
99.2
94
91.6
2004
2004
2004
2004
2004
2004
2004
2004
95.7
89.8
85.4
90.8
90.2
88.2
93.7
93.2
75.7
77.6
67.7
80.3
81.7
89.4
70.7
43.6
57
19
37.7
64.3
66
45.6
9.6
32.6
4.1
7.2
26.5
22.8
29.3
19.4
1.4
3.7
5.6
12.9
14.6
9
0.4
1.5
Males
50¬54
89.2
55¬59
76.4
97.3
88.2
60¬64
57.7
69.7 87.8
73.1
65¬69
43.0
66.1
47.5
70-74
25.0
88.6
52
54
49.3
Over 75 16.8
Over 80 7.4
34.6
Females
50¬54
37.9
40.9 19.9
18.6
55¬59
25.7
17.1
14.5
60¬64
16.9
26.3 13.4
13.0
65¬69
10.1
70-74
5.6
75-80
2.1
Over 80 1.3
Note:
8.7
10.2
13.3
70.3
57.3
28.1
30.2
6.1
68.4
52.1
74.5
77
76.4
81
78.7
59.6
34.8
65.0
56.2
61.7
70.9
70.1
39.7
12.6
45.4
16.2
19.7
52.7
38.9
24.0
1.9
23.3
2.5
4.1
19.1
10.1
14.9
12.0
0.7
1.5
3.6
5.4
5.8
4.3
0.2
0.5
2
8.6
Merged cells contain data for the combined age groups. For comparison purposes, labour force participa
rates for 65 plus and 75 plus age groups in Sri Lanka are, males - 27.53 and 12.02 and females 5.90 and 1
respectively
Source: ILO, LABORSTA Database
23
Table 4: Reasons for complete withdrawal from work of old workers
Very
Moderately
Not important
important
important
at all
Male
48
17
34
Health reasons
Female
47
23
30
Male
8
27
63
To do other things
(hobbies)
Female
27
31
40
Male
0
11
84
Family obligations
Female
3
12
78
Male
5
8
86
Disliked work
Female
9
12
74
37
18
43
Reached mandatory Male
retirement
Female
40
16
39
Male
7
17
73
Traveling to work was
stressful
Female
12
23
60
Male
7
5
81
Business closed
Female
11
3
74
Male
8
13
77
Work was too
stressful
Female
11
20
65
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
24
Don't
know
0
0
2
2
4
7
2
4
2
5
3
4
7
11
2
4
Total
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
Table 5: Descriptive statistics of variables used in multinomoial logit estimation
Completely Retired
382
Mean
Std.
Dev
Working part time
208
Mean
Std.
Dev
0.70
0.83
0.72
0.03
0.03
0.03
0.72
0.04
Individual
characteristics
Male
Sinhala
Head of Household
0.86
0.03
0.71
0.04
0.75
0.23
0.02
0.03
0.03
0.01
0.68
0.04
0.28
0.04
Marital status
Married
Previously married
Never married
0.04
0.02
0.12
0.23
0.26
0.21
0.17
0.02
0.03
0.03
0.02
0.02
0.14
0.03
0.36
0.22
0.04
0.03
0.17
0.03
Education
No schooling
Less than primary
Lower secondary
Passed O/Ls
Passed A/Ls
0.10
0.02
0.08
0.62
0.30
0.02
0.03
0.03
0.12
0.03
0.55
0.04
Health - 1
Very good health
Good health
Poor health
0.33
0.04
0.68
0.03
0.65
0.04
0.19
0.03
Wealth
Ever had chronic
illness
Receiving pension
Family Transfers
(proxy)
Adequate wealth
Own a house
Preparation
for retirement
Health - 2
Sources of
support
Pre-retirement
employment
status
Location
Sector
Full time
working
470
Mean
Std.
Dev
All
1060
Mean
Std.
Dev
0.75
0.86
0.76
0.02
0.02
0.02
0.72
0.85
0.73
0.02
0.01
0.02
0.74
0.24
0.02
0.02
0.02
0.01
0.73
0.24
0.02
0.02
0.02
0.01
0.11
0.39
0.37
0.11
0.02
0.02
0.03
0.03
0.02
0.01
0.12
0.33
0.30
0.16
0.09
0.01
0.02
0.02
0.01
0.01
0.12
0.58
0.30
0.02
0.03
0.03
0.11
0.59
0.31
0.01
0.02
0.02
0.56
0.03
0.62
0.02
0.04
0.79
0.01
0.02
0.19
0.79
0.01
0.02
0.24
0.65
0.02
0.03
0.29
0.60
0.02
0.02
0.04
0.23
0.73
0.01
0.02
0.02
0.04
0.27
0.69
0.01
0.02
0.02
0.03
0.07
0.30
0.01
0.02
0.03
0.26
0.13
0.21
0.02
0.01
0.01
0.60
0.03
0.40
0.02
0.38
0.78
0.03
0.03
0.80
0.03
0.35
0.56
0.03
0.03
0.32
0.04
0.55
0.04
Well
Prepared
Not
0.05
0.33
0.62
0.02
0.03
0.03
0.02
0.26
0.01
0.04
0.72
0.04
Public regular
Private regular
Casual/
contractual
Employer/Self
employed
Western province
0.51
0.19
0.10
0.03
0.03
0.02
0.33
0.04
0.15
0.03
0.22
0.03
0.20
0.02
0.31
0.04
0.40
0.03
0.35
0.04
0.33
0.03
0.36
0.02
Urban
0.15
0.02
0.13
0.02
0.11
0.02
0.13
0.01
Rural
0.75
0.03
0.79
0.03
0.87
0.02
0.81
0.01
Estate
0.10
0.02
0.08
0.02
0.02
0.01
0.06
0.01
Source: Own calculations using World Bank 2006 Sri Lanka Aging Survey.
25
Table 6: Multinomial logit estimates of transition from full-time employment1
Variable Description1
Marginal Effects
Working part time
Robust
std err
Coef2
Male
Individual
characteristics
Marital status
(currently
married)3
0.48
Sinhala
Health - 1 (very
good)
Health - 2 (no
chronic illness)
Sources of support
Preparation for
retirement (well)
Location
Sector (urban)
No obs.
Source:
0.30
-0.11
Working
part-time
Working
full-time
0.06
0.05
0.45
-0.42
0.38
0.02
0.12
-0.14
0.27
-0.04
0.26
0.01
-0.01
0.00
Previously married
0.60
0.29
0.41
0.30
-0.11
0.08
0.03
Never married
No schooling
0.97
0.70
0.90
0.74
-0.18
0.09
0.09
0.15
0.52
0.02
0.64
-0.02
0.03
-0.01
0.26
-0.30
0.44
0.45
0.27
0.16
0.61
0.59
-0.06
0.01
0.02
-0.07
0.04
0.07
Lower secondary
**
Passed O/Ls
0.01
0.42
0.13
0.61
-0.02
-0.01
0.03
Good health
-0.64
*
0.35
-0.96
**
0.35
0.18
-0.03
-0.15
Poor health
-0.84
**
0.41
-1.45
***
0.40
0.27
-0.04
-0.23
Ever had chronic
illness
Pension income
-0.13
0.25
-0.53
**
0.24
0.08
0.03
-0.11
0.34
-1.37
***
0.38
0.28
-0.07
-0.21
Some one to help
financially
Adequate wealth
0.04
0.28
-0.02
0.28
0.00
0.01
-0.01
0.23
0.27
0.11
0.27
-0.04
0.04
0.00
0.24
0.01
-0.05
0.04
-0.91
***
-0.23
0.24
0.08
Some
0.59
0.69
-0.03
0.85
-0.06
0.13
-0.07
Not
0.67
0.68
-0.24
0.83
-0.03
0.15
-0.12
-0.04
0.37
1.69
***
0.49
-0.22
-0.18
0.40
Casual/
contractual
Employer/Self
employed
Western province
0.83
**
0.37
3.66
***
0.47
-0.43
-0.23
0.66
0.55
*
0.33
3.72
***
0.45
-0.45
-0.24
0.69
-0.09
0.26
0.11
0.25
-0.01
-0.03
0.03
Rural
-0.10
0.29
0.26
0.37
-0.02
-0.05
0.07
Estate
-0.21
0.61
-1.32
0.57
0.17
0.06
-0.23
**
1060
Pseudo R2
Notes:
0.43
Completely
Retired
0.47
Private regular
Pre-retirement
employment status
(public regular)
0.29
Robust
std err
-0.09
Own a house
Wealth
*
Coef
Head of Household
Less than primary
Education
(A/levels or above)
Working full time
0.2144
1. Dependent variable takes three values : completely retiring upon reaching 61 years and not working when
aged 61-65 (taken as a baseline); working part time (at least some time during ages 61-65, either part time, or
full-time if not throughout the period, and working continuously full-time during ages 61-65.
2. Significance at 1, 5 and 10 per cent levels are indicated by ‘***’,’**’ and * respectively.
3. Omitted category is given in parenthesis.
Own calculations using World Bank 2006 Sri Lanka Aging Survey
26
Figure 1: Demographic aging, Sri Lanka and world regional
averages (percentage of population over 60 years old, 2000-2050)
35
S ri La nka
30
As ia
Percent
25
OEC D
20
15
La tin Am e ric a
a nd C a ribbe a n
10
Ea s te rn Euro pe
a nd fo rm e r S U
5
M iddle Ea s t a nd
No rth Afric a
0
2000
2010
2020
2030
2050
S ub-S a ha ra n
Afric a
Source: De Silva (2007), for Sri Lanka; World Bank (1994), for other countries
27
Figure 2: Trends in dependency ratios, Sri Lanka 2001-2101
100%
Percentage of population
93%
90%
Child dependency
80%
Old age dependency
76%
Total dependency
70%
60%
63%
54%
50%
50%
40%
37%
30%
29%
26%
20%
16%
10%
0%
2001
2011
2021
2031
2041
2051
Source: De Silva (2007)
28
2061
2071
2081
2091
2101
Figure 3: Labor force projections, 2001-2070
12000
10000
8000
standard
6000
low
4000
high
2000
20
01
20
07
20
13
20
19
20
25
20
31
20
37
20
43
20
49
20
55
20
61
20
67
0
Source: Calculated using population projections by De Silva (2007) and
labor force survey data by the Census and Statistics.
29
Figure 4: Age composition of standard labor force projections, by gender (2001-2070)
Females
Males
7000
3500
6000
3000
5000
2500
4000
2000
3000
1500
2000
1000
1000
500
0
20
01
20
06
20
11
20
16
20
21
20
26
20
31
20
36
20
41
20
46
20
51
20
56
20
61
20
66
0
Ye a r
Year
15- 29
30-39
40- 49
50- 59
60-69
70+
15-29
30-39
40-49
50-59
60-69
70+
Source: Calculated using population projections by de Silva (2007) and labor force survey data by the Census and
Statistics.
30
Figure 5: Labour force participation, by
age and gender, 1992 and 2004 (in percent)
120
100
80
60
40
20
0
15-19 20-29 30-39 40-49 50-59 60-69 70 +
Men 1992
Men 2004
Women 1992
Women 2004
Source: Own calculations based on various Labor Force Surveys of Sri Lanka.
31
Figure 6: Labor force participation of old
workers, 1992-2004 (percent)
60.0
40.0
20.0
Males 60-69
Males 70+
Females 60-69
Females 70+
20
04
20
02
20
00
19
98
19
96
19
94
19
92
0.0
Source: Own calculations based on various Labor Force Surveys of Sri Lanka.
32
Figure 7: Unemployment rates by age and gender
(1992 and 2004)
50
45
40
35
30
25
20
15
10
5
0
15-19
1992 40-49
20-29Men
30-39
Women 1992
Men2004
50-59
60-69 over 70
Women 2004
Source: Own calculations based on various Labor Force
Surveys of Sri Lanka.
33
Figure 8: Retirement status by age and gender (percent)
(a) Men
(b) Women
100
100
80
80
retired
%
60
%
work parttime
work fulltime
not relevant
40
20
0
60-64
65-69
70-74
Age group
75-80
retired
60
work part-time
40
work full-time
20
not relevant
60-64 65-69 70-74 75-80
80+
Age group
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
34
80+
Figure 9: Average number of hours worked, by age and gender
25
20
15
10
5
0
60-63
Men
Women
64-67
68-71
# of hours per week
Hours per week
(a) including fully retired, (b) employed workers only (full- and part-time workers)
72-75
50
40
30
M en
20
Women
10
0
60-63
64-67
Age
Age
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
35
68-71
72-75
Figure 10: Age wage profile, by sector and gender
Index (wages for 25-29 year olds = 100)
170
Wage index
150
130
110
90
70
50
30
2529
3034
3539
4044
4549
5054
5559
6064
65 +
Age group
Public sector male
Public sector female
Private sector male
Private sector female
Source: Own calculations based on various Labor Force
Surveys of Sri Lanka.
36
Figure 11: Old workers by type of employment status and type of employer
(a) By type of employment status
(b) By type of employer
70
60
40
%
%
50
30
20
10
0
60-64
70-74
Male
Regular
60-64
70-74
120
100
80
60
40
20
0
60-64
Casual
70-74
60-64
Male
Female
Self employed
Government
Formal private
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
37
70-74
Female
Informal private
Figure 12: Old workers by sector of work and occupation
45
40
35
30
25
20
15
10
5
0
(b) Old workers by occupation
%
%
(a) Old workers by sector of work
60-64
70-74
60-64
Male
Agriculture
70
60
50
40
30
20
10
0
60-64
70-74
Feamle
Manufacturing
70-74
Male
Services
Professionals
Skilled workers
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
38
60-64
70-74
Female
Elementary occupations
Figure 13: Retirement status, by prime-age employment type and age
(a) Regular public sector w orkers
(b) Regular private sector workers
100%
100%
80%
80%
60%
60%
40%
40%
20%
20%
0%
0%
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
Retired
Work part_time
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
Retired
Work full_time
Work part_time
Work full_time
(d) Self-employed
(c) Casual workers
100%
100%
80%
80%
60%
60%
40%
40%
20%
20%
0%
0%
55
56
57
58
Retired
59
60
61
62
63
64
65
66
Work part_time
67
68
69
70
71
72
73
74
55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
75
Retired
Work full_time
Work part_time
Work full_time
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
Note: The employment status refers to employment status at age 54. All workers were working full time between
ages 51 to 54.
Work related reasons: work stress, travel stress, mandatory retirement, completion of contract, business closed;
Health reasons: health reasons or illness, not feeling physically well.
Personal reasons: retirement incentives and family obligations.
39
Figure 14: Percentage of retirements by age groups
70
60
Publicsector
%
50
40
Private sector
30
Casual workers
20
10
Self_employed
0
55-59
60-64
65-69
70-75
76-80
80+
Age group
Source: Own calculations based on World Bank 2006 Sri
Lanka Aging Survey.
Note: Retirements from full-employment to either complete
retirement or part-time work.
40
Figure 15: Sector of work, by the prime-age employment type and age
(b) Regural private sector workers
(a) Regular public sector workers
100%
100%
80%
80%
60%
60%
40%
40%
`
20%
20%
age
Agriculture
Industry
Agriculture
Services
(c) Casual workers
75
73
71
69
67
65
age
Industry
Services
(d) Self-employed
20%
73
age
75
71
69
67
63
65
61
59
57
55
57
0%
73
75
40%
69
71
60%
63
65
67
80%
59
61
100%
80%
60%
40%
20%
0%
100%
55
63
59
61
55
75
73
71
69
67
65
63
61
59
57
55
57
0%
0%
age
Agriculture
Industry
Agriculture
Services
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
41
Industry
Services
Figure 16: Reasons for retirement from full time work
(a) Regular public sector workers
(b) Regular private sector workers
120
120
100
80
health
60
w rk_related
40
personal
%
%
100
20
80
health
60
w rk_related
40
personal
20
0
0
55
56-60
61-65
66-70
71-75
55
Age of re tire m e nt
health
w rk_related
personal
61-65
66-70
66-70
71-75
(d) Self-employed
71-75
%
%
(c ) Casual workers
56-60
61-65
age of Re tire m e nt
100
90
80
70
60
50
40
30
20
10
0
55
56-60
100
90
80
70
60
50
40
30
20
10
0
health
w rk_related
personal
55
age of Re tire m e nt
56-60
61-65
66-70
age of Re tire m e nt
Source: Own calculations based on World Bank 2006 Sri Lanka Aging Survey.
42
71-75
END NOTES
1
While modernization is affecting Asian societies in different ways, there is evidence of declining coresidence of
old people with their children in developed Asian countries -- Japan and South Korea (see Hermalin 1995, Knodel
and Ofstedal 2002, and Kim 1999), and also of a sharp decline in expected coresidence in Taiwan (Hermalin and
Young 2004).
2
The primary source for this analysis is a 2006 Aging Survey conducted in early 2006 by the World Bank to study
the impact of population aging in Sri Lanka. The total number of interviewed persons was 2413, selected randomly
among the population of 60 year olds and above. The sample was drawn based on pre-listing of old persons in
randomly selected 260 primary sampling units (North-East region was excluded). The questionnaire consisted of
several modules, with the present paper using the following ones: background and basic data, activities in which old
people have been engaged, income and assets, and health and health care use.
3
For most developed countries, projections show that their labor forces will stop growing in next decade unless
participation rates rise (OECD 2006).
4
These calculations assume constant future labor force participation rates and apply age and gender-specific labor
force participation rates (averages for 2000-04 period) to population projections. We also made projections
assuming higher female LFP rates (reflecting possible catching up with developed countries, as currently female
participation rates of Sri Lanka lag behind countries such as Japan, Malaysia and Thailand), showing that such
increases would delay the shrinking of labor force by a few years.
5
Simulations for developed countries show that the most powerful countervailing effects can be expected from the
increased labor force participation of old workers. Martins et al (2005) show over the long run, the impact of
increased retirement age in line with longevity gains prevails over impact of improved labor quality, and that this
effect is particularly large for countries currently having low participation rates of older workers. For example, an
increased labor force participation scenario in France would turn a decline in labor supply into an increase in labor
supply over the next four decades. In Japan, however, the intensity of aging is such that even the combined effect of
quality adjustments and increases in participation rates are sufficient to prevent an absolute decline in labor supply
by 2050. Similarly, projections for Eastern and Central European countries show that an increased labor force
participation of workers aged 40 to 59 years as well those over 60 years old significantly slows down the shrinking
of labor force (Muenz 2007).
6
The explanatory variables used in the regression include age, marital status, education, province, and number of
dependent children and old people in the household.
7
The question on retirement status was not relevant for 66 per cent of elderly females. This arises from the fact that
many females do not participate in market based economic activities in younger ages. According to DCS, Labor
Force Survey data 60 percent of females 60 to 64 years old are unavailable work, because they are engaged in
household activities. Although this percent decreases for elderly age-cohorts even at 80 plus years of age, 6% of
elderly females state that they are not available for work due to household work.
8
Note that these profiles are derived from cross-section data and may not necessarily reflect life-cycle profiles (that
is, age-wage profiles of the same workers traced through time), because these profiles may be heavily influenced by
the selection of workers: other things equal, high-paid workers are more likely to withdraw from the labor market
later in their career than low-paid workers. The increase of relative wages for workers in their early 60s compared
with workers in their late 50s in the private sector thus reflects two opposing effects: the selection effect (that is, a
higher likelihood that better paid workers stay employed) as well as the negative effect of age on wages of life-cycle
wage profile, with the former prevailing (that is, better-paid workers self-select and continue working while lowerpaid workers transfer to inactivity).
9
As described, for example, by OECD (2006), pull factors are associated primarily with financial incentives
provided by pension schemes and family obligations such as helping with childcare of grand children, and push
factors with firm and personal circumstances that restrict job opportunities of old workers (such as perceptions about
the capacities of old workers and mandated benefits or increases in wages associated with age or work experience).
10
Unpaid family workers were not included in the analysis, due to small (20 observations) sample size.
11
According to the pre-retirement employment status, the total sample of 1060 individuals is divided into 276
regular public sector workers, 138 regular private sector workers, 223 casual workers, and 424 self-employed (the
data are right censored).
12
A log rank test showed that the survival curves for the four sectors of employment – that is, regular public sector
workers, regular private sector workers, casual workers and self employed – were statistically different.
43
13
Sri Lanka’s severance pay system – embodied in the Termination of Employment of Workman Act (TEWA) of
1971 – is one of the costliest and most restrictive severance pay systems in the world. The TEWA requires
employers with more than 14 workers to seek the authorization of the Commissioner of Labor for intended layoffs.
It not only requires that employers pay high compensation to laid off workers, but its discretionary nature and
lengthy procedures further restrict the ability of employers to lay off workers. By ex-ante determining the amount of
severance pay, the recently introduced formulas somewhat reduced arbitrariness, but payments remain well above
international norms (for a description and consequences of TEWA, including on firms’ growth and job dynamics,
see World Bank 2007a).
14
For example, OECD (2006) reports that firms in Japan and Korea are massively resorting to mandatory retirement
age in order to evade strict employment protection legislation, because this legislation does not apply to mandatory
retirements.
15
Based on cross-country analysis, OECD (2006) provides some evidence of a negative impact of high relative
wages on employment opportunities of old workers (drop in retention and hiring rates of male older workers).
16
World Bank (2007b) points to the weaknesses of the current schemes for informal workers in Sri Lanka and lays
out an alternative strategy to expanding social security to informal sector workers.
44